Over the years, a number of semisupervised deep-learning algorithms have been proposed for time-series classification (TSC). In deep learning, from point view representation hierarchy, semantic information extracted lower levels is basis that higher levels. The authors wonder if high-level also helpful capturing low-level information. This paper studies this problem and proposes robust model with self-distillation (SD) simplifies existing learning (SSL) techniques TSC, called SelfMatch....

The existing multiobjective expected improvement (EI) criteria are often computationally expensive because they calculated using multivariate piecewise integrations, the number of which increases exponentially with objectives. In order to solve this problem, paper proposes a new approach develop cheap-to-evaluate EI based on proposed matrix (EIM). elements in EIM single-objective EIs that studying point has beyond each Pareto front approximation objective. Three developed by combining into...

Kriging models, also known as Gaussian process are widely used in surrogate-assisted evolutionary algorithms (SAEAs). However, the cubic time complexity of standard models limits their usage high-dimensional optimization. To tackle this problem, we propose an incremental model for computation. The main idea is to update incrementally based on equations previously trained instead building from scratch when new samples arrive, so that updating can be reduced quadratic. proposed learning scheme...

In mobile-edge computing (MEC), smart mobile devices (SMDs) with limited computation resources and battery lifetime can offload their computing-intensive tasks to MEC servers, thus enhance the capability reduce energy consumption of SMDs. Nevertheless, offloading edge incurs additional transmission time higher execution delay. This article studies tradeoff between completion applications SMDs in networks. The problem is formulated as a multiobjective (MCOP), where task precedence, i.e.,...

The Constrained Expected Improvement (CEI) criterion used in the so-called Efficient Global Optimization (C-EGO) algorithm is one of most famous infill criteria for expensive constrained optimization problems. However, standard CEI selects only point to evaluate each cycle, which time consuming when parallel computing architecture available. This work proposes a new Parallel EGO (PC-EGO) extend C-EGO computing. proposed PC-EGO tested on sixteen analytical problems as well real-world...

The undispatchability of wind farm presents lots integration problems to safe operation power system. Designing energy storage system for Dispatchability is an effective solution. This paper a novel sizing method dispatchability with different confidence level. forecast data generated base on persistence method, and then based dispatch strategy proposed the minimization size. For compromise between size, level can be defined as constraint function rated capacity established non-parametric...

The multipoint expected improvement (EI) criterion is a well-defined parallel infill for expensive optimization. However, the exact calculation of classical EI involves evaluating significant amount multivariate normal cumulative distribution functions, which makes inner optimization this very time consuming when number samples large. To tackle problem, we propose novel fast in work. proposed calculated using only univariate distributions; thus, it easier to implement and cheaper compute...

Abstract The blasting vibration produced in the process of underwater engineering brings serious damage to surrounding environment. Predicting peak particle velocity (PPV) is one effective ways alleviate problem. To further improve prediction accuracy blast PPV, grey wolf optimization (GWO) algorithm used this paper optimize penalty factor and radial basis kernel function parameters support vector regression (SVR) model iteratively, a PPV was established. Taking Dajin Island water intake...

Bayesian optimization (BO) algorithm is very popular for solving low-dimensional expensive problems. Extending to high dimension a meaningful but challenging task. One of the major challenges that it difficult find good infill solutions as acquisition functions are also high-dimensional. In this work, we propose expected coordinate improvement (ECI) criterion high-dimensional optimization. The proposed ECI measures potential can get by moving current best solution along one coordinate....

Extending Bayesian optimization to batch evaluation can enable the designer make most use of parallel computing technology. Most current approaches artificial functions simulate sequential algorithm's behavior select a points for evaluation. However, as size grows, accumulated error introduced by these increases rapidly, which dramatically decreases efficiency algorithm. In this work, we propose simple and efficient approach extend Different from existing approaches, idea new is draw...

The expected improvement (EI) criterion has been widely used in Kriging-assisted evolutionary algorithms to select individuals for expensive evaluations. It measures the amount of candidates are gain compared with current best solution, based on which will be picked Since all candidate measured by solution when calculating EI values, population moves gradually towards decrease diversity population. In this work, we propose a new anisotropic (AEI) resolve issue. Instead comparing proposed AEI...

Prescreening strategies have been widely used in surrogate-assisted evolutionary algorithms for screening out poor solutions. Existing prescreening are designed individual-level selection, i.e. they to select individuals from a set of population members. In this work, we propose strategy based on the multi-point expected improvement criterion Kriging-assisted algorithms. each generation proposed algorithm, operators repeatedly generate candidate populations. Then, these populations...